The dynamic nature of microtubules is critical for their function in cells. Microtubules are hollow tubes that polymerize by the addition of GTP-bound α/β-tubulin heterodimers to their ends. Tubulin incorporation induces GTP hydrolysis which destabilizes the microtubule wall. A delay in hydrolysis of unknown duration is thought to produce a 'GTP cap', a protective end structure formed of GTP-tubulins that is critical for microtubule stability (Brouhard and Rice, 2018; Carlier, 1982; Erickson and O'Brien, 1992; Howard and Hyman, 2009; Mitchison and Kirschner, 1984). The loss of this cap is thought to expose an unstable GDP lattice and trigger depolymerization, resulting in dynamic instability of microtubule growth (Mitchison and Kirschner, 1984). How the biochemistry of GTP hydrolysis translates into conformational changes in the growing microtubule end, and thereby determines the properties of the GTP cap and microtubule stability, is still not understood (Brouhard and Rice, 2018; Kueh and Mitchison, 2009).

End binding proteins of the EB family (EBs) bind autonomously to an extended region at growing microtubule ends where the GTP cap is expected to be (Bieling et al., 2008; Bieling et al., 2007; Dixit et al., 2009; Komarova et al., 2009; Maurer et al., 2012; Roth et al., 2018). The affinity of EBs for binding to the growing microtubule end region is about 10 fold higher than to the GDP part of the microtubule lattice at a distance from the end (Maurer et al., 2011; Seetapun et al., 2012). In cells, EBs control microtubule dynamics and function by locally recruiting numerous regulators of cytoskeleton function specifically to growing microtubule ends (Akhmanova and Steinmetz, 2015).

Whether EBs bind indeed to the GTP cap is still debated, mostly because they bind differently to microtubules grown in the presence of different non-hydrolysable GTP analogues. EBs bind to GTPγS microtubules with similar affinity than to growing microtubule ends, but considerably less to GMPCPP microtubules (Maurer et al., 2011; Roth et al., 2018). This is probably a consequence of structural differences between microtubules grown with these two GTP analogues (Alushin et al., 2014; Manka and Moores, 2018; Zhang et al., 2015; Zhang et al., 2018). However, in contrast to GTPγS, GMPCPP strongly promotes microtubule nucleation and stabilizes microtubules, and is therefore usually considered to be the better GTP analogue for microtubules (Alushin et al., 2014; Hyman et al., 1992; Manka and Moores, 2018). These observations with GTP analogues and the detection of a very short EB-free zone at the very end of growing microtubules, just in front of the EB binding region, have raised the question whether EBs really bind the GTP cap or possibly instead a post-hydrolysis state in the growing microtubule end region (Maurer et al., 2014; Zhang et al., 2015). It is therefore unclear whether the EB binding region and the GTP cap are identical and which GTP analogue induces more faithfully the GTP state in microtubules. This limits our understanding of the nature of the GTP cap.

Here we studied recombinant human microtubules with altered GTP hydrolysis kinetics to address these unanswered questions. We found that microtubules with blocked GTP hydrolysis were hyperstable, nucleated extremely efficiently and that EBs bound with high affinity along their entire length. Microtubules with impaired, but not blocked GTPase rate were also considerably more stable than wildtype microtubules and displayed higher nucleation efficiency. They had an elongated EB binding region at the growing microtubule ends, indicative of slowed-down GTP hydrolysis. Together, these results demonstrate that EBs do indeed bind with high affinity to the GTP conformation of the microtubule and consequently are GTP cap markers. Moreover, we observed an EB affinity gradient along the length of the GTP cap, indicative of a conformational and stability gradient within the cap as a result of GTP hydrolysis.

To examine the relationship between GTP hydrolysis and conformational changes in the GTP cap and to answer the question of the nucleotide preference of EBs, we produced recombinant human tubulin, which allowed us to alter the GTPase activity of microtubules. Human α/β-tubulin was expressed in insect cells and purified (and separated from endogenous insect cell tubulin) by affinity chromatography using an internal hexa-histidine-tag in α-tubulin (Sirajuddin et al., 2014; Vemu et al., 2016) and a protease-cleavable Strep-tag at the C-terminus of β-tubulin (Figure 1—figure supplement 1) We generated wildtype α/β-tubulin and a tubulin variant with blocked GTP hydrolysis by mutating the evolutionarily conserved catalytic glutamate 254 of α-tubulin (Nogales et al., 1998) to alanine (E254A) (Figure 1a–b, Figure 1—figure supplement 1). This mutation is lethal in budding yeast, as cell division fails due to loss of microtubule dynamics (Anders and Botstein, 2001).

Figure 1 with 2 supplements see all

Download asset Open asset

We measured the GTP content of microtubules polymerized from purified recombinant human wildtype tubulin and tubulin purified from porcine brain as a control. We observed the expected 1:1 ratio of GTP:GDP (Figure 1c–d), since GTP is hydrolyzed only at its inter-tubulin dimer, but not at the intra-dimer binding site (Kobayashi, 1975). In contrast, E254A microtubules bound almost exclusively GTP (Figure 1c–d), demonstrating that GTP hydrolysis is indeed blocked in this mutant.

To visualize the growth dynamics of label-free recombinant microtubules elongating from surface-attached stable microtubule 'seeds' (Figure 1—figure supplement 2a), we used interferometric scattering (iSCAT) microscopy (Ortega Arroyo et al., 2016). In the presence of GTP, human wildtype microtubules grew dynamically (Figure 1e–f, Video 1) similarly to porcine brain microtubules (Figure 1—figure supplement 2b–c), displaying occasional transitions to depolymerization, called catastrophes, as expected (Vemu et al., 2017; Vemu et al., 2016).

Video 1

Download asset

Wildtype tubulin hardly nucleated any microtubules spontaneously (Figure 1g, Videos 1 and 2), whereas E254A tubulin nucleated microtubules very efficiently even at low concentrations, quickly reaching high microtubule densities (Figure 1h, Video 3). Moreover, these mutant microtubules were hyper-stable for hours, never exhibiting catastrophes. These properties agree with the anticipated behavior of hydrolysis-deficient microtubules, and are similar to the behavior of microtubules polymerized in the presence of the non-hydrolysable GTP analogue GMPCPP (Hyman et al., 1992).

Video 2

Download asset

Video 3

Download asset

Next, we tested how EBs bind to human wildtype and to hydrolysis-deficient E254A microtubules. Simultaneous total internal reflection fluorescence (TIRF) microscopy of mGFP-EB3 (Figure 2—figure supplement 1a) and iSCAT imaging of unlabeled microtubules revealed that EB3 bound to the growing plus and minus ends of wildtype microtubules in the expected comet-like manner (Figure 2a–c, Video 4; Bieling et al., 2008; Bieling et al., 2007; Komarova et al., 2009). The apparent dissociation constant for binding to growing microtubule ends was ~40 nM (Figure 2d). EB3 had no effect on the wildtype microtubule growth speed (Figure 2e). 

In clear contrast, EB3 decorated the entire lattice of GTPase-deficient E254A mutant microtubules (Figure 2f–h; Figure 2—figure supplement 1b), reminiscent of its binding to microtubules grown in the presence of the non-hydrolysable GTP analogue GTPγS (Maurer et al., 2011). However, EB3 failed to bind human wildtype microtubules polymerized in the presence of GMPCPP (Figure 2—figure supplement 1c), as observed previously with mammalian brain microtubules (Maurer et al., 2011; Roth et al., 2018). This negative effect of GMPCPP on the EB3 binding affinity was also observed when E254A microtubules were grown in the presence of this GTP analogue (Figure 2—figure supplement 1d), demonstrating that it is indeed the bound nucleotide and not the single point mutation itself that controls the microtubule conformation and hence the affinity of EB3 binding. With an apparent dissociation constant of ~8 nM (Figure 2i), EB3 binding to E254A microtubules was stronger than to growing wildtype ends (Figure 2d). Unbinding of mGFP-EB3 from E254A microtubules was very slow (Figure 2j–k, Figure 2—figure supplement 1f–g). Together these results show that EB3 displays strong affinity to microtubules locked in the GTP state, clearly establishing it as a bona fide GTP cap marker.

Figure 2 with 1 supplement see all

Download asset Open asset

Video 4

Download asset

Interestingly, E254A microtubules simultaneously capture properties that different GTP analogues induce separately in wildtype microtubules: strong microtubule nucleation as in the presence of GMPCPP, and high EB binding affinity as in the presence of GTPγS (Hyman et al., 1992; Maurer et al., 2011). Another protein that discriminates between a GMPCPP microtubule and a growing microtubule end is a short fragment of the multifunctional anti-catastrophe factor TPX2 (Wittmann et al., 2000), previously called TPX2^micro (Zhang et al., 2017). Opposite to EBs, it displays higher affinity towards GMPCPP microtubules than towards growing microtubule ends (Zhang et al., 2017). We observed that this fragment also binds poorly to E254A microtubules compared to GMPCPP microtubules (Figure 2—figure supplement 1e), again suggesting that GTP-bound mutant microtubules are similar to growing microtubule ends. These results suggest that microtubules grown in GTP analogues display only partial aspects of the GTP conformation of microtubules, which explains previous contradictory interpretations based on results obtained with such analogues. In conclusion, a GTP-containing E254A mutant microtubule currently appears to be the best static mimic for the microtubule conformation in the only transiently existing GTP cap.

To investigate how GTP hydrolysis determines conformational transformations within the GTP cap, we generated a human tubulin mutant, in which glutamate 254 is substituted by aspartate (E254D) (Figure 3a). This chemically similar but shorter catalytic residue is expected to slow down GTP hydrolysis compared to wildtype microtubules. E254D microtubules grew steadily from surface-immobilized seeds (Figure 3b) and EB3 tracked their growing ends, displaying strikingly longer and brighter fluorescent 'comets' (Figure 3c–d, Videos 5–7), clearly indicative of slower GTP hydrolysis. E254D microtubules hardly ever displayed any catastrophes even under conditions where wildtype microtubules exhibited frequent depolymerization events (Figure 3e–f, Videos 5–7). E254D tubulin also nucleated microtubules more efficiently in solution than wildtype tubulin, but less than E254A tubulin (Figure 3b, Figure 1h) indicating an intermediate stability of E254D microtubules compared to wildtype and completely GTPase-deficient E254A microtubules. This suggests that GTP hydrolysis is indeed slowed down in the E254D mutant.

Figure 3

Download asset Open asset

Video 5

Download asset

Video 6

Download asset

Video 7

Download asset

E254D microtubules grew about twofold faster than wildtype microtubules, either with or without EB3 (Figure 3h). A comparison of the average EB3 intensity profiles along the ends of wildtype and E254D microtubules growing at tubulin concentrations adjusted such that they grow with comparable speeds, shows that around ~5 times more EB3 binds to the ends of E254D microtubules (Figure 3i). This increase can to a large extent be explained by the longer EB3 binding region on the mutant microtubules (Figure 3j). Quantitative analysis of the intensity profiles allows to extract the characteristic comet lengths and, knowing the measured growth speeds, also the lifetime of the EB3 binding region (Bieling et al., 2008; Bieling et al., 2007; Maurer et al., 2014; Maurer et al., 2012), which was four-fold longer at the ends of E254D microtubules compared to wildtype microtubules (Figure 3j). These results demonstrate that slowed-down GTP hydrolysis extends the lifetime and hence the size of the GTP cap.

We occasionally observed 'split comets' at the growing ends of E254D microtubules that could later re-join (Figure 3g, magenta arrowheads) or detach and grow individually (Figure 3g, cyan arrowheads). This indicates that even partial microtubule end structures can be more stable when GTP hydrolysis is slow. This is reminiscent of recently observed split microtubule ends in the presence of the protofilament capping drug eribulin and the microtubule-stabilizing protein CLASP (Aher et al., 2018; Doodhi et al., 2016). The ‘curved’ appearance of these partial comets likely reflects global conformational differences between unfinished end structures such as observed for microtubule nucleation and fully formed tubes, pointing to a high degree of structural plasticity of the GTP cap region (Roostalu and Surrey, 2017; Voter and Erickson, 1984; Brouhard and Rice, 2018; Igaev and Grubmüller, 2018).

How the GTP cap stabilizes growing microtubule ends is an open question. As GTP is hydrolyzed over time, a nucleotide state gradient may form at the growing microtubule end that in turn might translate into a conformational stability gradient. Such a gradient would be detectable as a gradient of EB affinities within the GTP cap. We therefore probed the microtubule conformation along the length of the GTP cap by measuring the binding strength of EB3 at different positions within the GTP cap using single molecule imaging and dwell time analysis (Figure 4—figure supplement 1). We first took advantage of the elongated GTP caps of growing E254D microtubules.

We imaged single mGFP-EB3 molecules binding in the elongated cap region of growing E254D microtubules, with their end regions labeled by excess Alexa647-EB3 (Figure 4a). We observed an overall complex (non-mono-exponential) distribution of mGFP-EB3 dwell times in the GTP cap, indicative of the presence of different conformational states with different EB3 binding affinities in the cap (Figure 4b). Next, we performed a spatially resolved dwell time analysis by generating several 'local dwell time distributions' for distinct distances from the growing microtubule end. These local distributions were rather mono-exponential, indicative of a more homogenous local conformational lattice state (Figure 4c). The local characteristic EB3 dwell times decreased along the length of the GTP cap (Figure 4c–d, Figure 4—figure supplement 2a). This indicates a decreasing EB binding affinity along the length of the GTP cap, suggesting a gradual conformational change within the cap. The long EB3 dwell times at the very tip of the GTP cap are indicative of a GTP-like state, given that EB3 binds strongly to GTPase deficient E254A microtubules (Figure 2f–k), which then transforms into the lower affinity GDP state concomitant with GTP hydrolysis. The conformational change along the length of the GTP cap most likely reflects a stability gradient whose characteristic length corresponds to the cap length (Figure 4d).

Figure 4 with 3 supplements see all

Download asset Open asset

We observed a similar gradual conformational change in the shorter GTP caps of wildtype microtubules growing at a similar speed (Figure 4—figure supplement 3), when we recorded a considerably larger number of single molecule binding events compared to previous studies (Bieling et al., 2008; Bieling et al., 2007; Dixit et al., 2009; Maurer et al., 2011; Maurer et al., 2014) to enable spatially resolved dwell time analysis (Figure 4e–h). We detected also here a combination of conformational states within the whole GTP cap, as indicated by a complex EB3 dwell time distribution for all events observed in the entire cap region (Figure 4f), whereas rather homogenous conformational states were detected at distinct positions within the GTP cap, as indicated by more mono-exponential local dwell time distributions (Figure 4g). The mean dwell times, indicative of the EB3 binding affinity, decreased over a shorter distance in wildtype GTP caps (Figure 4h, Figure 4—figure supplement 2c) than in E254D caps (Figure 4d, Figure 4—figure supplement 2a), reflecting the shorter EB3 comet length at wildtype microtubule ends compared to longer EB3 comets at E254D microtubule ends with reduced GTPase activity (Figure 3j, Figure 4—figure supplement 2b,d). Taken together, these results suggest that a conformational gradient in the GTP cap at the growing microtubule end is a response to the GTP hydrolysis in the lattice.

Using recombinant mutated human tubulin we produced for the first time pure GTP-microtubules, eliminating the need for using non-hydrolysable GTP analogues. GTP-microtubules were hyper-stable, as predicted by the GTP cap hypothesis, and EBs bound strongly to these GTP microtubules. Microtubules with impaired GTP hydrolysis (E254D) were also more stable than wildtype microtubules and displayed longer EB comets, indicative of longer GTP caps. Our results are to date the most direct demonstration that the kinetics of GTP hydrolysis determine GTP cap length and that EBs bind to the GTP cap, explaining why their binding region represents the part of the microtubule that stabilizes it (Duellberg et al., 2016; Maurer et al., 2012).

The length of the GTP cap of recombinant human wildtype microtubules was in a similar range as previously measured EB 'comet' lengths with mammalian brain microtubules in vitro (Bieling et al., 2008; Bieling et al., 2007; Dixit et al., 2009; Maurer et al., 2011; Rickman et al., 2017; Roth et al., 2018; Vemu et al., 2017) and in cells (Seetapun et al., 2012). This confirms that recombinant human wildtype tubulin and mammalian brain tubulin hydrolyze GTP with similar speeds (Vemu et al., 2017). The hydrolysis rate appears to be set such that sufficiently long GTP caps are created to stabilize growing microtubules during fluctuating growth (Duellberg et al., 2016; Gardner et al., 2011; Rickman et al., 2017; Schek et al., 2007).

Our results suggest that previous observations with GTP analogues require careful interpretation, because neither GMPCPP nor GTPγS generate microtubules with properties identical to microtubules that contain GTP. When considering microtubule stability and the affinity for EB binding, each analogue appears to produce some, but not all features of the microtubule lattice conformation in the GTP cap.

The recent observation of a short EB-free region at the very end of growing microtubules (Maurer et al., 2014) together with the observation of EBs not binding GMPCPP very well, had stimulated a discussion about the possibility of EBs binding a post-hydrolysis and not a GTP induced conformation of the microtubule lattice (Maurer et al., 2014; Zhang et al., 2015). Given the results presented here, this can now be considered unlikely. The observed short EB-free zone at the extreme microtubule end is probably rather a consequence of finite EB association kinetics (Maurer et al., 2014) and of some particular structural feature at the very end of growing microtubules. EB binding sites that are located between protofilaments (Maurer et al., 2012; Zhang et al., 2015), may not form there due to a flared growing microtubule end structure lacking lateral protofilaments interactions, as observed recently (McIntosh et al., 2018), or another GTP lattice conformation may exist at the very microtubule end.

For wildtype microtubules and microtubules with slowed-down GTP hydrolysis, we performed here for the first time spatially resolved dwell time measurements of single EB3 molecules at different positions of the GTP cap. Our aim was to gain insight into the conformational changes of the microtubule lattice within the cap that affect the conformation of the EB binding site, allowing it to monitor the lattice conformation. The binding strength of EB3 decreased gradually from the microtubule end towards the GDP lattice (Figure 4d,h), both for wildtype and E254D caps. Binding strength decayed mono-exponentially on a length scale similar to the comet lengths measured by 'comet' analysis (Figure 3j), suggesting that the density of EBs decreases along the length of the cap due to an affinity decrease. The associated time scale of this affinity decrease as the microtubule grows provides an estimate for the GTP hydrolysis rate which is ~0.2 s⁻¹ for wildtype microtubules and four times slower for E254D microtubules.

Using a mean squared displacement (MSD) analysis of single EB3 molecules bound to E254D caps we find that EB3 diffuses on the microtubule lattice with a diffusion coefficient of ~10⁻³ µm²/s. Although this is several magnitudes slower than EB1 lattice diffusion reported previously under different conditions on microtubules grown in nucleotide analogues (Lopez and Valentine, 2016), it means that during a typical binding event EB3 samples the conformational state of several tubulins in a range of up to ~50 nm in our experiments. This defines the spatial resolution of this conformational reporter. We can not discriminate if all tubulins in the sampled area have exactly the same conformation or if EBs report on the average conformation of a lattice with individual tubulins being in different conformational states. Nevertheless, EBs clearly detect a conformational gradient within the GTP cap, as evidenced by the dependence of its mean dwell time on the distance from the growing microtubule end.

Interestingly, the ratio between the dwell times of EB3 bound to E254A microtubules (Figure 2h–i, Figure 2—figure supplement 1f,g) versus wildtype microtubule ends (Figure 4f) was considerably larger than the ratio between the corresponding equilibrium association constants (inverse equilibrium dissociation constants) (Figure 2d and i). This indicates that not only the unbinding rate, but also the binding rate of EB3 is reduced on E254A microtubules compared to wildtype microtubules, even if so to a lesser extent. This suggests that EB3 may induce and stabilize a conformational lattice transformation in E254A microtubules when it binds, akin to an 'induced fit' scenario. Such an effect on the microtubule lattice would agree with previous observations showing that EB binding to microtubules can change their protofilament twist (von Loeffelholz et al., 2017; Zhang et al., 2015; Zhang et al., 2018) and mildly accelerate cap maturation (Maurer et al., 2011). We can now conclude that the latter means 'accelerate GTP hydrolysis', in agreement with our results here as well as recent structural observations (Zhang et al., 2015).

In conclusion, our results demonstrate that the kinetics of GTP hydrolysis critically determine the properties of the GTP cap the size of which can be monitored by EBs. Because EBs serve as a binding platform for numerous microtubule binding proteins it ensures that microtubule dynamics in cells is regulated at multiple levels by targeting the most critical modulator of microtubule stability, the GTP cap, itself. Future challenges will be to elucidate the high-resolution structures of microtubules in their true GTP state and of the conformational transitions within the GTP cap to decipher the determinants of microtubule stability and function at atomic resolution.

All data generated or analysed during this study are included in the manuscript and supporting files.

1. 1. Aher A
   2. Kok M
   3. Sharma A
   4. Rai A
   5. Olieric N
   6. Rodriguez-Garcia R
   7. Katrukha EA
   8. Weinert T
   9. Olieric V
   10. Kapitein LC
   11. Steinmetz MO
   12. Dogterom M
   13. Akhmanova A

   (2018) CLASP suppresses microtubule catastrophes through a single TOG domain

   Developmental Cell 46:40–58.

   https://doi.org/10.1016/j.devcel.2018.05.032

   • PubMed
   • Google Scholar
2. 1. Akhmanova A
   2. Steinmetz MO

   (2015) Control of microtubule organization and dynamics: two ends in the limelight

   Nature Reviews Molecular Cell Biology 16:711–726.

   https://doi.org/10.1038/nrm4084

   • PubMed
   • Google Scholar
3. 1. Alushin GM
   2. Lander GC
   3. Kellogg EH
   4. Zhang R
   5. Baker D
   6. Nogales E

   (2014) High-Resolution Microtubule Structures Reveal the Structural Transitions in αβ-Tubulin upon GTP Hydrolysis

   Cell 157:1117–1129.

   https://doi.org/10.1016/j.cell.2014.03.053

   • Google Scholar
4. 1. Anders KR
   2. Botstein D

   (2001) Dominant-lethal alpha-tubulin mutants defective in Microtubule depolymerization in yeast

   Molecular Biology of the Cell 12:3973–3986.

   https://doi.org/10.1091/mbc.12.12.3973

   • PubMed
   • Google Scholar
5. 1. Bieling P
   2. Laan L
   3. Schek H
   4. Munteanu EL
   5. Sandblad L
   6. Dogterom M
   7. Brunner D
   8. Surrey T

   (2007) Reconstitution of a microtubule plus-end tracking system in vitro

   Nature 450:1100–1105.

   https://doi.org/10.1038/nature06386

   • PubMed
   • Google Scholar
6. 1. Bieling P
   2. Kandels-Lewis S
   3. Telley IA
   4. van Dijk J
   5. Janke C
   6. Surrey T

   (2008) CLIP-170 tracks growing microtubule ends by dynamically recognizing composite EB1/tubulin-binding sites

   The Journal of Cell Biology 183:1223–1233.

   https://doi.org/10.1083/jcb.200809190

   • PubMed
   • Google Scholar
7. 1. Bieling P
   2. Telley IA
   3. Hentrich C
   4. Piehler J
   5. Surrey T

   (2010) Fluorescence microscopy assays on chemically functionalized surfaces for quantitative imaging of microtubule, motor, and +TIP dynamics

   Methods Cell Biology 95:555–580.

   https://doi.org/10.1016/S0091-679X(10)95028-0

   • Google Scholar
8. 1. Bohner G
   2. Gustafsson N
   3. Cade NI
   4. Maurer SP
   5. Griffin LD
   6. Surrey T

   (2016) Important factors determining the nanoscale tracking precision of dynamic microtubule ends

   Journal of Microscopy 261:67–78.

   https://doi.org/10.1111/jmi.12316

   • PubMed
   • Google Scholar
9. 1. Brouhard GJ
   2. Rice LM

   (2018) Microtubule dynamics: an interplay of biochemistry and mechanics

   Nature Reviews Molecular Cell Biology 19:451–463.

   https://doi.org/10.1038/s41580-018-0009-y

   • PubMed
   • Google Scholar
10. 1. Carlier MF

    (1982) Guanosine-5'-triphosphate hydrolysis and tubulin polymerization. Review article

    Molecular and Cellular Biochemistry 47:97–113.

    https://doi.org/10.1007/BF00234410

    • PubMed
    • Google Scholar
11. 1. Castoldi M
    2. Popov AV

    (2003) Purification of brain tubulin through two cycles of polymerization-depolymerization in a high-molarity buffer

    Protein Expression and Purification 32:83–88.

    https://doi.org/10.1016/S1046-5928(03)00218-3

    • PubMed
    • Google Scholar
12. 1. Dixit R
    2. Barnett B
    3. Lazarus JE
    4. Tokito M
    5. Goldman YE
    6. Holzbaur EL

    (2009) Microtubule plus-end tracking by CLIP-170 requires EB1

    PNAS 106:492–497.

    https://doi.org/10.1073/pnas.0807614106

    • PubMed
    • Google Scholar
13. 1. Doodhi H
    2. Prota AE
    3. Rodríguez-García R
    4. Xiao H
    5. Custar DW
    6. Bargsten K
    7. Katrukha EA
    8. Hilbert M
    9. Hua S
    10. Jiang K
    11. Grigoriev I
    12. Yang CH
    13. Cox D
    14. Horwitz SB
    15. Kapitein LC
    16. Akhmanova A
    17. Steinmetz MO

    (2016) Termination of protofilament elongation by eribulin induces lattice defects that promote microtubule catastrophes

    Current Biology 26:1713–1721.

    https://doi.org/10.1016/j.cub.2016.04.053

    • PubMed
    • Google Scholar
14. 1. Duellberg C
    2. Cade NI
    3. Holmes D
    4. Surrey T

    (2016) The size of the EB cap determines instantaneous microtubule stability

    eLife 5:e13470.

    https://doi.org/10.7554/eLife.13470

    • PubMed
    • Google Scholar
15. 1. Erickson HP
    2. O'Brien ET

    (1992) Microtubule dynamic instability and GTP hydrolysis

    Annual Review of Biophysics and Biomolecular Structure 21:145–166.

    https://doi.org/10.1146/annurev.bb.21.060192.001045

    • PubMed
    • Google Scholar
16. 1. Gardner MK
    2. Charlebois BD
    3. Jánosi IM
    4. Howard J
    5. Hunt AJ
    6. Odde DJ

    (2011) Rapid microtubule self-assembly kinetics

    Cell 146:582–592.

    https://doi.org/10.1016/j.cell.2011.06.053

    • PubMed
    • Google Scholar
17. 1. Gebhardt JC
    2. Suter DM
    3. Roy R
    4. Zhao ZW
    5. Chapman AR
    6. Basu S
    7. Maniatis T
    8. Xie XS

    (2013) Single-molecule imaging of transcription factor binding to DNA in live mammalian cells

    Nature Methods 10:421–426.

    https://doi.org/10.1038/nmeth.2411

    • PubMed
    • Google Scholar
18. 1. Howard J
    2. Hyman AA

    (2009) Growth, fluctuation and switching at Microtubule plus ends

    Nature Reviews Molecular Cell Biology 10:569–574.

    https://doi.org/10.1038/nrm2713

    • PubMed
    • Google Scholar
19. 1. Hyman A
    2. Drechsel D
    3. Kellogg D
    4. Salser S
    5. Sawin K
    6. Steffen P
    7. Wordeman L
    8. Mitchison T

    (1991) Preparation of modified tubulins

    Methods in Enzymology 196:478–485.

    https://doi.org/10.1016/0076-6879(91)96041-o

    • PubMed
    • Google Scholar
20. 1. Hyman AA
    2. Salser S
    3. Drechsel DN
    4. Unwin N
    5. Mitchison TJ

    (1992) Role of GTP hydrolysis in Microtubule dynamics: information from a slowly hydrolyzable analogue, GMPCPP

    Molecular Biology of the Cell 3:1155–1167.

    https://doi.org/10.1091/mbc.3.10.1155

    • PubMed
    • Google Scholar
21. 1. Igaev M
    2. Grubmüller H

    (2018) Microtubule assembly governed by tubulin allosteric gain in flexibility and lattice induced fit

    eLife 7:e34353.

    https://doi.org/10.7554/eLife.34353

    • PubMed
    • Google Scholar
22. 1. Jha R
    2. Roostalu J
    3. Cade NI
    4. Trokter M
    5. Surrey T

    (2017) Combinatorial regulation of the balance between dynein microtubule end accumulation and initiation of directed motility

    The EMBO Journal 36:3387–3404.

    https://doi.org/10.15252/embj.201797077

    • PubMed
    • Google Scholar
23. 1. Kobayashi T

    (1975)

    Dephosphorylation of tubulin-bound guanosine triphosphate during microtubule assembly

    Journal of Biochemistry 77:1193–1197.

    • PubMed
    • Google Scholar
24. 1. Komarova Y
    2. De Groot CO
    3. Grigoriev I
    4. Gouveia SM
    5. Munteanu EL
    6. Schober JM
    7. Honnappa S
    8. Buey RM
    9. Hoogenraad CC
    10. Dogterom M
    11. Borisy GG
    12. Steinmetz MO
    13. Akhmanova A

    (2009) Mammalian end binding proteins control persistent microtubule growth

    The Journal of Cell Biology 184:691–706.

    https://doi.org/10.1083/jcb.200807179

    • PubMed
    • Google Scholar
25. 1. Kueh HY
    2. Mitchison TJ

    (2009) Structural plasticity in actin and tubulin polymer dynamics

    Science 325:960–963.

    https://doi.org/10.1126/science.1168823

    • PubMed
    • Google Scholar
26. 1. Lopez BJ
    2. Valentine MT

    (2016) The +TIP coordinating protein EB1 is highly dynamic and diffusive on microtubules, sensitive to GTP analog, ionic strength, and EB1 concentration

    Cytoskeleton 73:23–34.

    https://doi.org/10.1002/cm.21267

    • PubMed
    • Google Scholar
27. 1. Manka SW
    2. Moores CA

    (2018) The role of tubulin-tubulin lattice contacts in the mechanism of microtubule dynamic instability

    Nature Structural & Molecular Biology 25:607–615.

    https://doi.org/10.1038/s41594-018-0087-8

    • PubMed
    • Google Scholar
28. 1. Maurer SP
    2. Bieling P
    3. Cope J
    4. Hoenger A
    5. Surrey T

    (2011) GTPgammaS microtubules mimic the growing microtubule end structure recognized by end-binding proteins (EBs)

    PNAS 108:3988–3993.

    https://doi.org/10.1073/pnas.1014758108

    • PubMed
    • Google Scholar
29. 1. Maurer SP
    2. Fourniol FJ
    3. Bohner G
    4. Moores CA
    5. Surrey T

    (2012) EBs recognize a nucleotide-dependent structural cap at growing microtubule ends

    Cell 149:371–382.

    https://doi.org/10.1016/j.cell.2012.02.049

    • PubMed
    • Google Scholar
30. 1. Maurer SP
    2. Cade NI
    3. Bohner G
    4. Gustafsson N
    5. Boutant E
    6. Surrey T

    (2014) EB1 accelerates two conformational transitions important for microtubule maturation and dynamics

    Current Biology 24:372–384.

    https://doi.org/10.1016/j.cub.2013.12.042

    • PubMed
    • Google Scholar
31. 1. McIntosh JR
    2. O'Toole E
    3. Morgan G
    4. Austin J
    5. Ulyanov E
    6. Ataullakhanov F
    7. Gudimchuk N

    (2018) Microtubules grow by the addition of bent guanosine triphosphate tubulin to the tips of curved protofilaments

    The Journal of Cell Biology 217:2691–2708.

    https://doi.org/10.1083/jcb.201802138

    • PubMed
    • Google Scholar
32. 1. Mitchison T
    2. Kirschner M

    (1984) Dynamic instability of microtubule growth

    Nature 312:237–242.

    https://doi.org/10.1038/312237a0

    • PubMed
    • Google Scholar
33. 1. Montenegro Gouveia S
    2. Leslie K
    3. Kapitein LC
    4. Buey RM
    5. Grigoriev I
    6. Wagenbach M
    7. Smal I
    8. Meijering E
    9. Hoogenraad CC
    10. Wordeman L
    11. Steinmetz MO
    12. Akhmanova A

    (2010) In vitro reconstitution of the functional interplay between MCAK and EB3 at Microtubule plus ends

    Current Biology 20:1717–1722.

    https://doi.org/10.1016/j.cub.2010.08.020

    • PubMed
    • Google Scholar
34. 1. Nogales E
    2. Downing KH
    3. Amos LA
    4. Löwe J

    (1998) Tubulin and FtsZ form a distinct family of GTPases

    Nature Structural Biology 5:451–458.

    https://doi.org/10.1038/nsb0698-451

    • PubMed
    • Google Scholar
35. 1. Ortega Arroyo J
    2. Cole D
    3. Kukura P

    (2016) Interferometric scattering microscopy and its combination with single-molecule fluorescence imaging

    Nature Protocols 11:617–633.

    https://doi.org/10.1038/nprot.2016.022

    • PubMed
    • Google Scholar
36. 1. Rickman J
    2. Duellberg C
    3. Cade NI
    4. Griffin LD
    5. Surrey T

    (2017) Steady-state EB cap size fluctuations are determined by stochastic microtubule growth and maturation

    PNAS 114:3427–3432.

    https://doi.org/10.1073/pnas.1620274114

    • PubMed
    • Google Scholar
37. 1. Roostalu J
    2. Cade NI
    3. Surrey T

    (2015) Complementary activities of TPX2 and chTOG constitute an efficient importin-regulated microtubule nucleation module

    Nature Cell Biology 17:1422–1434.

    https://doi.org/10.1038/ncb3241

    • PubMed
    • Google Scholar
38. 1. Roostalu J
    2. Surrey T

    (2017) Microtubule nucleation: beyond the template

    Nature Reviews Molecular Cell Biology 18:702–710.

    https://doi.org/10.1038/nrm.2017.75

    • PubMed
    • Google Scholar
39. 1. Roth D
    2. Fitton BP
    3. Chmel NP
    4. Wasiluk N
    5. Straube A

    (2018) Spatial positioning of EB family proteins at Microtubule tips involves distinct nucleotide-dependent binding properties

    Journal of Cell Science 132:jcs219550.

    https://doi.org/10.1242/jcs.219550

    • PubMed
    • Google Scholar
40. 1. Ruhnow F
    2. Zwicker D
    3. Diez S

    (2011) Tracking single particles and elongated filaments with nanometer precision

    Biophysical Journal 100:2820–2828.

    https://doi.org/10.1016/j.bpj.2011.04.023

    • PubMed
    • Google Scholar
41. 1. Schek HT
    2. Gardner MK
    3. Cheng J
    4. Odde DJ
    5. Hunt AJ

    (2007) Microtubule assembly dynamics at the nanoscale

    Current Biology 17:1445–1455.

    https://doi.org/10.1016/j.cub.2007.07.011

    • PubMed
    • Google Scholar
42. 1. Seetapun D
    2. Castle BT
    3. McIntyre AJ
    4. Tran PT
    5. Odde DJ

    (2012) Estimating the microtubule GTP cap size in vivo

    Current Biology 22:1681–1687.

    https://doi.org/10.1016/j.cub.2012.06.068

    • PubMed
    • Google Scholar
43. 1. Sirajuddin M
    2. Rice LM
    3. Vale RD

    (2014) Regulation of microtubule motors by tubulin isotypes and post-translational modifications

    Nature Cell Biology 16:335–344.

    https://doi.org/10.1038/ncb2920

    • PubMed
    • Google Scholar
44. 1. Snapp EL
    2. Hegde RS
    3. Francolini M
    4. Lombardo F
    5. Colombo S
    6. Pedrazzini E
    7. Borgese N
    8. Lippincott-Schwartz J

    (2003) Formation of stacked ER cisternae by low affinity protein interactions

    The Journal of Cell Biology 163:257–269.

    https://doi.org/10.1083/jcb.200306020

    • PubMed
    • Google Scholar
45. 1. Ti SC
    2. Alushin GM
    3. Kapoor TM

    (2018) Human beta-Tubulin isotypes can regulate microtubule protofilament number and stability

    Developmental Cell 47:175–190.

    https://doi.org/10.1016/j.devcel.2018.08.014

    • PubMed
    • Google Scholar
46. 1. Vemu A
    2. Atherton J
    3. Spector JO
    4. Szyk A
    5. Moores CA
    6. Roll-Mecak A

    (2016) Structure and dynamics of Single-isoform recombinant neuronal human tubulin

    Journal of Biological Chemistry 291:12907–12915.

    https://doi.org/10.1074/jbc.C116.731133

    • PubMed
    • Google Scholar
47. 1. Vemu A
    2. Atherton J
    3. Spector JO
    4. Moores CA
    5. Roll-Mecak A

    (2017) Tubulin isoform composition tunes microtubule dynamics

    Molecular Biology of the Cell 28:3564–3572.

    https://doi.org/10.1091/mbc.e17-02-0124

    • PubMed
    • Google Scholar
48. 1. von Loeffelholz O
    2. Venables NA
    3. Drummond DR
    4. Katsuki M
    5. Cross R
    6. Moores CA

    (2017) Nucleotide- and Mal3-dependent changes in fission yeast microtubules suggest a structural plasticity view of dynamics

    Nature Communications 8:2110.

    https://doi.org/10.1038/s41467-017-02241-5

    • PubMed
    • Google Scholar
49. 1. Voter WA
    2. Erickson HP

    (1984)

    The kinetics of microtubule assembly. Evidence for a two-stage nucleation mechanism

    The Journal of Biological Chemistry 259:10430–10438.

    • PubMed
    • Google Scholar
50. 1. Wasilko DJ
    2. Lee SE
    3. Stutzman-Engwall KJ
    4. Reitz BA
    5. Emmons TL
    6. Mathis KJ
    7. Bienkowski MJ
    8. Tomasselli AG
    9. Fischer HD

    (2009) The titerless infected-cells preservation and scale-up (TIPS) method for large-scale production of NO-sensitive human soluble guanylate cyclase (sGC) from insect cells infected with recombinant baculovirus

    Protein Expression and Purification 65:122–132.

    https://doi.org/10.1016/j.pep.2009.01.002

    • PubMed
    • Google Scholar
51. 1. Wittmann T
    2. Wilm M
    3. Karsenti E
    4. Vernos I

    (2000) TPX2, A novel xenopus MAP involved in spindle pole organization

    The Journal of Cell Biology 149:1405–1418.

    https://doi.org/10.1083/jcb.149.7.1405

    • PubMed
    • Google Scholar
52. 1. Zacharias DA
    2. Violin JD
    3. Newton AC
    4. Tsien RY

    (2002) Partitioning of lipid-modified monomeric GFPs into membrane microdomains of live cells

    Science 296:913–916.

    https://doi.org/10.1126/science.1068539

    • PubMed
    • Google Scholar
53. 1. Zhang R
    2. Alushin GM
    3. Brown A
    4. Nogales E

    (2015) Mechanistic origin of microtubule dynamic instability and its modulation by EB proteins

    Cell 162:849–859.

    https://doi.org/10.1016/j.cell.2015.07.012

    • PubMed
    • Google Scholar
54. 1. Zhang R
    2. Roostalu J
    3. Surrey T
    4. Nogales E

    (2017) Structural insight into TPX2-stimulated microtubule assembly

    eLife 6:e30959.

    https://doi.org/10.7554/eLife.30959

    • PubMed
    • Google Scholar
55. 1. Zhang R
    2. LaFrance B
    3. Nogales E

    (2018) Separating the effects of nucleotide and EB binding on microtubule structure

    PNAS 115:E6191–E6200.

    https://doi.org/10.1073/pnas.1802637115

    • PubMed
    • Google Scholar

Acceptance summary:

This carefully executed in vitro reconstitution study resolves some important questions concerning the nature of the cap that stabilizes growing microtubules. The study will be of interest to a broad biological audience interested in cytoskeletal regulation.

Decision letter after peer review:

Thank you for submitting your article "The speed of GTP hydrolysis determines GTP cap size and controls microtubule stability" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by Anna Akhmanova as the Senior and Reviewing Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

The reviewers agreed that this is a rigorously performed study, which addresses important questions about the relationship between the stabilizing cap at growing microtubule ends, the nucleotide state of tubulin and the tubulin state recognized by End Binding proteins. The finding that EB3 specifically recognizes the GTP form of tubulin is an important, though perhaps not entirely surprising contribution to the field. Further, all three reviewers agreed that the description of a gradient of EB binding affinity along the microtubule end is the most novel and interesting part of the paper, but also the part that needs to be worked out better. Specifically, the conformational transition model raised a number of critiques, and these should be addressed by additional experiments and analyses,and possibly also by adjusting the writing of the paper. Since there might be different ways of extending this part of the paper, I include below a summary of the reviewers' discussions and also the full reviews.

The reviewers felt that you should be more clear about what you mean by 'conformation' – do you mean 'expanded/compacted' microtubule lattices, as described by structural work from Nogales lab? Could GTP tubulin exist in either the expanded or the compacted state according to a conformational equilibrium? To begin to address this question, it would be great to get data on whether E254A/D lattices are expanded or compacted. It should be possible to obtain these data without generating a high resolution structure, perhaps by using TPX2 binding experiments or by a 2D cryo-EM analysis.

Further, the conclusions about a conformational gradient are based on a limited set of data, and as, pointed out by the reviewers, alternative explanations of the obtained results are possible. One could consider performing experiments with a monomeric EB3 protein, because they might help to rule out or provide support for the model proposed by reviewer 1. Another idea raised during the consultation between the reviewers was to use as a binding substrate microtubules grown from mixtures of E254A and wild type tubulin at different ratios. Two reviewers questioned the assignment of 'mono exponential' distributions in Figure 4, and this critique should be addressed. Finally, two reviewers found that the split comets deserve some analysis, and the faster growth rate displayed by E254D microtubules requires some attention.

Reviewer #1:

Roostalu et al. have successfully created GTPase-dead and GTP-slow recombinant human tubulin, which is a significant technical breakthrough. They use this new tubulin, along with fluorescent-EB3, to probe the behavior the GTP cap. They make the surprising observation that the dwell time of EB3 molecules changes as you go deeper in the cap. As expected from the Surrey lab, this is a high quality paper addressing a central issue in the microtubule field, namely the role of GTP hydrolysis.

Major comments:

1) The dwell time of EB3 on the microtubule decreases as you move deeper into the cap. The decrease in dwell times is hypothesized to be caused by a gradual "conformational gradient" that reduces EB3 affinity. The visual representation of this idea is the color gradient from red to yellow in the schematic in Figure 4J.

I have an alternative hypothesis for why the dwell times decrease. Consider that EB3 is an "obligatory dimer" (Sen et al., 2013), so its dissociation requires both CH domains to unbind from their respective binding sites. At the very end of the microtubule, EB3 is likely to find two adjacent sites where all of the relevant tubulin dimers are in the GTP state. (The site has 4 dimers, but we can leave aside for the moment the question of their relative relevance.) As you go deeper into the cap, the binding sites will become a mixture of GTP, GDP-Pi, and GDP states. These states have different affinities for EB3. In some cases, a single, high-affinity, GTP site may be adjacent only to GDP sites. In the context of the EB3 dimer, the site mismatch means that one CH domain will bind a GTP site and one can only bind a low-affinity GDP site. EB3 becomes functionally monomeric, hanging on to the microtubule with only one hand. In contrast, EB3 at the very end of the microtubule is holding on with both hands. Note: the EB3 construct they are using appears to be full-length when I trace back through their Materials and methods references.

Can this alternative framework explain the decrease in dwell times with depth inside the cap (Figure 4C)? More specifically: a gradual transition from dimeric to monomeric affinity conditions, driven by the probabilistic availability to two binding sites with different affinities? Are the measurements precise enough to rule out this hypothesis? The visual representation of this idea would NOT be a color gradient but rather an increased "speckling" of red and yellow blocks.

2) I have some technical questions about the "spatially-resolved EB3 dwell time distributions". They divide the microtubules into bins based on distance from the plus-end. The bin size is approximately 0.6 μm for the E254D microtubules (see legend to Figure 4C, see Figure 4—figure supplement 2) and 0.2 μm for the wildtype microtubules (Figure 4—figure supplement 2). These bins are relatively small when you consider that: (1) the microtubule end-position is not determined with sub-pixel accuracy, but is determined rather using the "traced end position" from a manual-tracing of a kymograph, which they state has 0.2 μm accuracy, (2) there is error in the single-molecule localization of the EB3 of 30 nm, and (3) the microtubule is growing at 70 nm/s, which means that an EB3 molecule could start its residence time in one bin and yet end it in another. The first two points make me uncertain whether the molecules are being correctly placed into their bins. The 3rd point is more conceptual: what is the best way to treat a molecule whose bin changes beneath it?

3) How mono-exponential are the spatially-resolved dwell time distributions (Figure 4C)? They appear to start off linear but then deviate from linearity at, e.g., t = 4 s for the 1.47 μm bin.

The central argument of the paper hinges on the fact that Figure 4B is not mono-exponential but Figure 4C is mono-exponential. Are the fits really strong enough to support this? The distributions are described as "strikingly mono-exponential". What does strikingly mean in terms of goodness of fit?

4) The split-comet phenomenon is mentioned briefly, along with their "curved appearance" – but there is very little quantification of the behavior. How often are they observed? How bright are the split comets compared to a full comet, etc.

5) The tubulin that they use retains an internal His tag on the α subunit. That's fine; the Roll-Mecak lab uses similar constructs. The Materials and methods are clear about the retention of the internal His tag but the main text is not. I think it's important to be clear throughout.

A suggestion:

The E254A and E254D microtubules are characterized in terms of their EB3 affinities. Perhaps the authors could explore the binding preference of TPX2, as it would report on the expansion/compaction state of these lattices? Alternatively, in the concluding paragraph, the authors say that high-resolution structures are on the way. But one doesn't necessarily need a 3.4 A structure to answer the most important question: are these lattices expanded or compacted?

Notes on the writing:

– The hypothesis that a nucleotide gradient "might translate into a conformational stability gradient" comes out of nowhere. What motivates this hypothesis? Are there data, structures, kinetic measurements, conceptual arguments, etc, that would motivate this idea?

– The concluding paragraph suggests that the "normal" GTP hydrolysis rate might be under evolutionary pressure. Are there measurements of the GTPase activity with non-human proteins that would provide support for this idea?

Reviewer #2:

This is an interesting and well-executed paper that addresses interesting questions about the microtubule's stabilizing cap, how it relates to nucleotide state, and what is the preferred state that EB proteins recognize. The approach was to purify site-directed recombinant human α/β-tubulins with mutations at a putative GTPase residue. The experiments are performed rigorously and for the most part described clearly, and the work has been done to a high technical standard.

The main findings of the paper are: (i) that abolishing (or at least substantially slowing; E254A) GTPase results in very stable microtubules akin to GMPCPP microtubules, (ii) that moderately slowing (E254D) GTPase results in microtubules with longer EB comets that are also more stable, and (iii) that in these longer comets (and also wild-type comets), there appears to be a gradient of EB binding affinity, with the highest affinity being closest to the growing end. Although it has not been shown directly before using a mutant, I did not find it all surprising that reducing GTPase rate increased microtubule stability, but it is nice to see this in the way that it is shown here. The findings involving EB binding are more interesting: they provide some of the most direct support for the idea of an adaptable microtubule lattice, and they raise questions about what states commonly used GTP analogs are giving.

Major comments:

– I wonder if the balance of the manuscript should change somewhat if to be published in eLife. I think the site-specific EB analysis is probably the most interesting and mechanistic part of the manuscript, and the authors might want to place a little more emphasis there (and place less emphasis on some of the obvious-sounding claims). I don't think new experiments are required, but they may want to go deeper into some of the analysis of the EB binding.

– The authors observe that EB3 does not bind to recombinant GMPCPP microtubules, even though in other ways GMPCPP has been taken to be a good GTP analog. The authors did not mention that GMPCPP generally gives 14 protofilament microtubules. Could the authors commend on whether they think the switch in protofilament number or the different conformation of tubulin (or both) account for the lack of EB3 binding?

– Some of the language suggests there is a 1:1 correspondance between nucleotide state and lattice state, but their results indicate the opposite. I think this should be made more clear.

– It seems E254D microtubules grow ~twice as fast as wild-type. This is unexpected, and the authors really don't say much about it.

– In Figure 4D, the authors fit the EB dwell time decay as a function of distance from the tip. This is what they measured, of course. Since they know the growth rates, is there anything interesting to learn from plotting the decay against time (or just transforming to get the time dependence)? It seems that even speculatively they might be able to get a little more specific about the likely mix of GTP/GDP at different distances from the tip, and/or say something more quantitative about what the GTPase rate must be.

– The spatially resolved EB binding analysis was a strength of the paper for me. I think it also has the potential to be confusing to people, because the language/fits imply a series of discrete 'states' but it seems these states would have to be heterogenous in terms of nucleotide state. A few more sentences about this might be useful.

– The supplemental discussion felt looser than the rest of the paper. In particular, while the authors ascribe various deficiencies to nucleotide analogs, they do not seem to consider the possibility that the mutation(s) they made might also perturb the structure. This criticism applies to the main text also. Some version of the discussion of induced fit should probably be incorporated into the main text.

Reviewer #3:

The study by Roostalu et al. addresses the fundamental question of how GTP-tubulin at the growing tip of a microtubule can be detected by other proteins. The present study examines the binding of the major plus end tip tracking protein, EB3, and its ability to specifically recognize the GTP form of tubulin. In recent years, it has been controversial as to whether tip trackers are recognizing GTP or possibly another nucleotide/conformational state, such as GDP-Pi-tubulin. Here, it is shown, using tubulin mutants that either fail to hydrolyze or hydrolyze slowly, that EB3 specifically recognizes the GTP form of tubulin. This is an important finding for the field, one that is achieved through elegant experimental studies of mutant tubulin and careful quantitative analysis, for which the authors are to be commended. However, in the final analysis, the authors invoke a GDP-Pi intermediate state, without strong evidence to justify it. Rather, it seems possible that a simpler alternative explanation that only assumes GTP and GDP states, as suggested by their data, is not ruled out. Thus, I am concerned that the study, while making an important contribution to the field, may be misleading in its final interpretation.

Major comments:

1) Conclusion of a GDP-Pi state based on non-exponentially distributed dwell times is problematic.

a) Figure 4C. The spatially resolved dwell time is interesting, but the two positions that are farther away from the tip appear that they may be bi-exponential. It seems the non-exponential distribution might be expected as the k_off jumps when the hydrolysis occurs, which would give spatially varying dwell times and nonexponential distributions as the hydrolysis can occur at random (Poisson process) during the observation time. So it still seems that the data could be consistent with a simple GTP-GDP model, with no need for an additional third state of GDP-Pi.

b) Figure 4J is unconvincing as the simpler model of only two nucleotide states (GTP and GDP) has not been ruled out. To rule this out it will be necessary to do model-convolution to make it convincing that the analysis method is not yielding spurious results. Even then, the EB binding could be dependent on the local neighborhood of nucleotide state (see Seetapun et al., Figure 2—figure supplement 1 for conformational spread modeling), which seems reasonable since the binding site is at the interface between tubulins. Overall, model-convolution on the microtubule addition-loss-hydrolysis and EB binding-unbinding to assess the model is needed to rule out the simpler GTP-GDP model. Even then the argument for a GDP-Pi state is not compelling.

2) The values for KDs of 8 nM and 40 nM for EB3 binding to GTP- and GDP-tubulin, respectively, seem very strong compared to Seetapun et al. values of 3.8 µM and 55 µM, respectively, in vivo. Why is this, and are the in vitro results informative of the tip tracking in living cells?

3) How big is the cap with wildtype tubulin in vitro? Note: "cap size" is in the title, but it is not estimated, despite a lot of nice quantitation. Previous estimates put it at ~40 in vitro (at 5 µM wildtype tubulin) and ~750 in vivo (Schek et al., 2006; Seetapun et al., 2012), a disparity that is largely explained by the relative disparity in the net growth rates under these two conditions. These papers should be cited, as previous estimates of cap size. (Note: need to account for tubulin concentration, e.g. 19 µM wildtype tubulin in Figure 3J vs. 5 µM in Schek et al. and 7 µM in. Seetapun et al.).

The reviewers agreed that this is a rigorously performed study, which addresses important questions about the relationship between the stabilizing cap at growing microtubule ends, the nucleotide state of tubulin and the tubulin state recognized by End Binding proteins. The finding that EB3 specifically recognizes the GTP form of tubulin is an important, though perhaps not entirely surprising contribution to the field.

We thank the reviewers for their overall very positive evaluation of the quality, novelty and importance of our work. We however respectfully disagree with the view that the demonstration of EBs binding GTP microtubules is less significant, because it is perceived as not surprising. The question whether EBs bind to the GTP conformation of microtubules could not be answered with certainty in the past. Previous studies have used various nucleotide analogs to address questions of the nucleotide state. However, EBs bind with quite different affinities to microtubules grown in either GMPCPP or GTPγS, which has left room for different interpretations and resulted in lack of clarity. In our Introduction, we now provide more background and explain these different interpretations that arose from partly contradictory observations. We believe that this helps to provide a better context and highlights the value of our experiments with GTP microtubules. In our view, we present here for the first time evidence that EBs indeed bind the GTP state of microtubules with high affinity, providing clarity concerning a central question about microtubule biochemistry that has remained unsolved and has been a matter of speculation.

Further, all three reviewers agreed that the description of a gradient of EB binding affinity along the microtubule end is the most novel and interesting part of the paper, but also the part that needs to be worked out better. Specifically, the conformational transition model raised a number of critiques, and these should be addressed by additional experiments and analyses,and possibly also by adjusting the writing of the paper. Since there might be different ways of extending this part of the paper, I include below a summary of the reviewers' discussions and also the full reviews.

We have performed additional analysis (going beyond what the reviewers asked) which helped us to improve the clarity of the description of the measured affinity gradient. We describe this in detail in our response to the individual points raised by the reviewers and provide an improved description and discussion in the manuscript.

The reviewers felt that you should be more clear about what you mean by 'conformation' – do you mean 'expanded/compacted' microtubule lattices, as described by structural work from Nogales lab? Could GTP tubulin exist in either the expanded or the compacted state according to a conformational equilibrium? To begin to address this question, it would be great to get data on whether E254A/D lattices are expanded or compacted. It should be possible to obtain these data without generating a high resolution structure, perhaps by using TPX2 binding experiments or by a 2D cryo-EM analysis.

EBs sense the conformation of their binding site. And the conformation of this binding site is affected by the nucleotide state. We make this clear in the revised Discussion. Fluorescence microscopy experiments cannot visualise the degree of lattice 'compaction' or 'expansion' that can be observed by cryo-electron microscopy. Therefore, we do not make statements about such lattice characteristics. Following a suggestion of a reviewer, we added experiments with a fragment of TPX2 to the manuscript but consider electron microscopy experiments beyond the scope of this already extensive study.

Further, the conclusions about a conformational gradient are based on a limited set of data, and as, pointed out by the reviewers, alternative explanations of the obtained results are possible. One could consider performing experiments with a monomeric EB3 protein, because they might help to rule out or provide support for the model proposed by reviewer 1. Another idea raised during the consultation between the reviewers was to use as a binding substrate microtubules grown from mixtures of E254A and wild type tubulin at different ratios. Two reviewers questioned the assignment of 'mono exponential' distributions in Figure 4, and this critique should be addressed. Finally, two reviewers found that the split comets deserve some analysis, and the faster growth rate displayed by E254D microtubules requires some attention.

We respectfully disagree with the view that our data set is too limited. Not many labs are currently able to make high quality recombinant tubulin. Although it cannot be produced in amounts as large as for animal brain tubulin, our single molecule imaging data sets here are larger than previous data sets, even when compared to experiments made with brain tubulin. Otherwise our new spatially resolved dwell time analysis at growing microtubule ends would not have been possible. We consider experiments with monomeric EB and mixed microtubule lattices beyond scope, because they have their own challenges and become easily studies in their own right. We have addressed the issue of the 'mono-exponential distributions' by additional analysis. We agree that the increased growth speed of the E254D mutant is interesting as well as the observation of ‘split comets’, which supports our conclusions about enhanced microtubule stability. However, we cannot characterize and address all the aspects that are interesting about these new mutants in such great detail (structure, kinetics, ‘split comets’, binding of different MAPs) in a single paper.

Reviewer #1:

[…]

1) The dwell time of EB3 on the microtubule decreases as you move deeper into the cap. The decrease in dwell times is hypothesized to be caused by a gradual "conformational gradient" that reduces EB3 affinity. The visual representation of this idea is the color gradient from red to yellow in the schematic in Figure 4J.

I have an alternative hypothesis for why the dwell times decrease. Consider that EB3 is an "obligatory dimer" (Sen et al., 2013), so its dissociation requires both CH domains to unbind from their respective binding sites. At the very end of the microtubule, EB3 is likely to find two adjacent sites where all of the relevant tubulin dimers are in the GTP state. (The site has 4 dimers, but we can leave aside for the moment the question of their relative relevance.) As you go deeper into the cap, the binding sites will become a mixture of GTP, GDP-Pi, and GDP states. These states have different affinities for EB3. In some cases, a single, high-affinity, GTP site may be adjacent only to GDP sites. In the context of the EB3 dimer, the site mismatch means that one CH domain will bind a GTP site and one can only bind a low-affinity GDP site. EB3 becomes functionally monomeric, hanging on to the microtubule with only one hand. In contrast, EB3 at the very end of the microtubule is holding on with both hands. Note: the EB3 construct they are using appears to be full-length when I trace back through their Materials and methods references.

Can this alternative framework explain the decrease in dwell times with depth inside the cap (Figure 4C)? More specifically: a gradual transition from dimeric to monomeric affinity conditions, driven by the probabilistic availability to two binding sites with different affinities? Are the measurements precise enough to rule out this hypothesis? The visual representation of this idea would NOT be a color gradient but rather an increased "speckling" of red and yellow blocks.

This is an interesting point and it turns out that it is important to consider whether EBs diffuse on the microtubule lattice while they are bound:

If they do not diffuse, the reviewer's model is not supported by the data. As pointed out by the reviewer, in the case of a "speckled" lattice (some GTP, some GDP, some GDP+Pi tubulins), one expects many binding affinities for a dimeric EB (which we use here). This would then result in complicated, clearly non-mono-exponential local dwell time distributions, which we do not observe.

However, if EBs can diffuse on the lattice, as reported earlier (Lopez and Valentine, 2016), they are expected to have a "mixed" affinity resulting from the various affinities of the binding sites they visit during lattice diffusion, leading again to mono-exponential local dwell time distributions. In our experiments, we observe much less diffusion than reported in Lopez and Valentine (possibly due to our lower ionic strength buffer). Performing a mean-squared-displacement (MSD) analysis of the positional variation of single EBs on E254D microtubules, we find a diffusion coefficient of D=8.4 x10^-4 um²/s. This means that during a typical dwell time of a few seconds, EBs appear to diffuse a distance of ~50 nm, corresponding to several tubulins.

Therefore, our data do not allow to distinguish between the reviewer's model of a speckled nucleotide state lattice in the cap of growing microtubule ends and a model in which tubulins hypothetically change their conformation in a concerted manner.

Nevertheless, we clearly observe an affinity gradient for EB binding at growing microtubule ends, as we state in our manuscript. We explain now in our Discussion that the measured affinities are 'average' affinities integrating information of the conformation of several EB binding sites visited by EBs during lattice diffusion. We also clarify that our colour gradient in the Legend of the schematic figure is intended to illustrate affinity gradient for EB binding.

2) I have some technical questions about the "spatially-resolved EB3 dwell time distributions". They divide the microtubules into bins based on distance from the plus-end. The bin size is approximately 0.6 μm for the E254D microtubules (see legend to Figure 4C, see Figure 4—figure supplement 2A) and 0.2 μm for the wildtype microtubules (Figure 4—figure supplement 2C). These bins are relatively small when you consider that: (1) the microtubule end-position is not determined with sub-pixel accuracy, but is determined rather using the "traced end position" from a manual-tracing of a kymograph, which they state has 0.2 μm accuracy, (2) there is error in the single-molecule localization of the EB3 of 30 nm, and (3) the microtubule is growing at 70 nm/s, which means that an EB3 molecule could start its residence time in one bin and yet end it in another. The first two points make me uncertain whether the molecules are being correctly placed into their bins. The 3rd point is more conceptual: what is the best way to treat a molecule whose bin changes beneath it?

A bin size of 200 nm and 600 nm was chosen for wildtype and E254D microtubules that grew at ~60 nm/s in experiments used for spatially resolved dwell time analysis. Figures 4F and 4B show that >90% of binding events have dwell times <4s (wildtype) or <6s (E254D); hence, only a very small fraction of EBs will be at a distance from the microtubule end that corresponds then to a different bin at the end compared to the beginning of the binding event. Mis-assignment to incorrect bins due the errors in determining the microtubule end position and the EB position, could lead to incorrectly mixing up different affinities in a bin which would make dwell time distributions less mono-exponential. However, this is not observed, therefore this is not an issue here. We do observe that total dwell time distributions which include all binding events are clearly less mono-exponential (Figure 4B) than the local dwell time distributions (Figure 4C). Importantly, the measured dwell times of the local dwell time distributions quite obviously change with distance from the microtubule end (Figure 4C), demonstrating the existence of an affinity gradient and providing a very simple explanation for the non-mono-exponential nature of the dwell time distribution of all events.

3) How mono-exponential are the spatially-resolved dwell time distributions (Figure 4C)? They appear to start off linear but then deviate from linearity at, e.g., t = 4 s for the 1.47 μm bin.

The central argument of the paper hinges on the fact that Figure 4B is not mono-exponential but Figure 4C is mono-exponential. Are the fits really strong enough to support this? The distributions are described as "strikingly mono-exponential". What does strikingly mean in terms of goodness of fit?

Our main conclusion of the spatially resolved dwell time analysis is that one can measure an affinity gradient for EB binding in the cap at growing microtubule ends. This is demonstrated by a change of the characteristic local dwell times with distance from the microtubule end. This result is independent of the question how well distributions are fitted by a mono-exponential distribution. However, we have demonstrated this now quantitatively: we state in the legend of Figure 4 that a mono-exponential fit to the total dwell time distribution (Figure 4B) has a reduced chi-squared value two-times larger than that of a global mono-exponential fit to all the distance-binned data (Figure 4—figure supplement 2A). Furthermore, it should be noted that the logarithmic display of the dwell time distributions overemphasizes the small deviations from mono-exponential behaviour in the tail of some distributions – there is remarkably little deviation over the largest decade of data.

4) The split-comet phenomenon is mentioned briefly, along with their "curved appearance" – but there is very little quantification of the behavior. How often are they observed? How bright are the split comets compared to a full comet, etc.

We clarify in the revised manuscript that comets with curved appearance occur for the E254D mutant at higher tubulin concentrations (8 µM or above), i.e. when microtubule growth speeds are fast and when considerable microtubule nucleation is observed. We found it interesting to report the 'split comets' (as previously reported in Portran et al., NCB, 2017) as they suggest that even partial tubes can grow under some conditions if caps are hyperstabilised and long. This is however not the main point of the manuscript. This could be characterized further in the future combined with structural studies.

5) The tubulin that they use retains an internal His tag on the α subunit. That's fine; the Roll-Mecak lab uses similar constructs. The Materials and methods are clear about the retention of the internal His tag but the main text is not. I think it's important to be clear throughout.

We agree. We state now in the main text that we use an internal His-tag. We already stated in the Materials and methods of the original version of the manuscript that an N-terminal tag on α-tubulin, even if it is very small, did not allow normal microtubule growth in our hands. We observed that highly purified recombinant wildtype tubulin that had 3 additional amino acids left at the N-terminus of α-tubulin after protease cleavage of an N-terminal His tag did not elongate from stabilised microtubule ‘seeds’. Therefore, we moved the His-tag to the position in α-tubulin that has been previously used by the Roll-Mecak, Vale and Rice labs for human and yeast tubulins. This allows normal microtubule growth.

A suggestion:

The E254A and E254D microtubules are characterized in terms of their EB3 affinities. Perhaps the authors could explore the binding preference of TPX2, as it would report on the expansion/compaction state of these lattices? Alternatively, in the concluding paragraph, the authors say that high-resolution structures are on the way. But one doesn't necessarily need a 3.4 A structure to answer the most important question: are these lattices expanded or compacted?

We agree that electron microscopy of these mutant microtubules is very interesting, but we consider this beyond the scope of this manuscript. Instead, we have made the suggested TPX2 experiment and include it in the revised version of the manuscript (Figure 2—figure supplement 1E). We used a fragment of TPX2 that we previously called 'TPX2^micro' and that binds preferentially to GMPCPP microtubules, but not very efficiently to growing microtubule ends (Zhang et al., 2017). We find that this construct does not bind well to GTP-containing E254A microtubules, demonstrating that the TPX2^micro binding site conformation is similar in the cap of normal growing microtubule ends and all along GTPase-deficient microtubules, but different on GMPCPP microtubules. Together with our findings for the EB binding preferences, this means that GTP-containing E254A microtubules mimic better the properties of normal GTP caps than GMPCPP microtubules, at least as far as EB3 and TPX2^micro binding is concerned.

Notes on the writing:

– The hypothesis that a nucleotide gradient "might translate into a conformational stability gradient" comes out of nowhere. What motivates this hypothesis? Are there data, structures, kinetic measurements, conceptual arguments, etc, that would motivate this idea?

The idea of a stability gradient is supported by our data of GTPase-dead microtubules (E254A) being extremely stable and nucleating extremely efficiently, and the microtubules with slow GTPase hydrolysis having an intermediate stability and nucleation capacity, in between E254A and wildtype microtubules. This allows the hypothesis that a conformational gradient in the GTP cap corresponds to a stability gradient of the lattice as it transforms from the GTP to the GDP state.

– The concluding paragraph suggests that the "normal" GTP hydrolysis rate might be under evolutionary pressure. Are there measurements of the GTPase activity with non-human proteins that would provide support for this idea?

We removed the mentioning of 'evolutionary pressure'.

Reviewer #2:

[…]

– I wonder if the balance of the manuscript should change somewhat if to be published in eLife. I think the site-specific EB analysis is probably the most interesting and mechanistic part of the manuscript, and the authors might want to place a little more emphasis there (and place less emphasis on some of the obvious-sounding claims). I don't think new experiments are required, but they may want to go deeper into some of the analysis of the EB binding.

We have expanded our Introduction to better explain the conflicting interpretations in the literature with regards to nucleotide state and microtubule stability. We believe that these contradictions necessitate the experiments that we report here with pure GTP microtubules in order to be able to clearly answer a fundamental questions about the nature of the GTP cap. We think that all our findings and not only the final part of the study provide important new information and clarify unresolved issues about the GTP cap and microtubule dynamics.

– The authors observe that EB3 does not bind to recombinant GMPCPP microtubules, even though in other ways GMPCPP has been taken to be a good GTP analog. The authors did not mention that GMPCPP generally gives 14 protofilament microtubules. Could the authors commend on whether they think the switch in protofilament number or the different conformation of tubulin (or both) account for the lack of EB3 binding?

Although EBs have been shown to promote the formation of 13-protofilament microtubules (Vitre et al., NCB, 2008; Maurer et al., 2011; Howes et al., Cell Cycle, 2018), they do also bind to microtubules with other protofilament numbers. There is typically a mixture of microtubules with different protofilament numbers in in vitro experiments. Nevertheless, there are no strong differences in EB binding between different microtubules neither at growing microtubule ends, nor on GTPγS microtubules to which EBs bind with high affinity (e.g. Maurer et al., 2011; Roth et al., 2018)) indicating that EBs can bind well also to 14 protofilament microtubule ends and 14 protofilament GTPγS microtubules. This demonstrates that the nucleotide state plays the dominant role when considering the behaviour of EBs on GMPCPP microtubules.

– Some of the language suggests there is a 1:1 correspondance between nucleotide state and lattice state, but their results indicate the opposite. I think this should be made more clear.

Our data do not allow us to decide whether there is a 1:1 correspondence between nucleotide and lattice state of an individual tubulin. However, we observe a correlation between EB binding affinity and nucleotide state. EB binding probably samples the average lattice state of the area it can diffuse over, as we explain in our new Discussion.

– It seems E254D microtubules grow ~twice as fast as wild-type. This is unexpected, and the authors really don't say much about it.

We agree that this is unexpected. We cannot explain this observation at the moment and expect that a combination of careful kinetic and structural experiments will be required to obtain a satisfactory mechanistic explanation. We consider this a study in itself and a task for the future.

– In Figure 4D, the authors fit the EB dwell time decay as a function of distance from the tip. This is what they measured, of course. Since they know the growth rates, is there anything interesting to learn from plotting the decay against time (or just transforming to get the time dependence)? It seems that even speculatively they might be able to get a little more specific about the likely mix of GTP/GDP at different distances from the tip, and/or say something more quantitative about what the GTPase rate must be.

The rate of the decay of the dwell times over distance together with the growth speed provides the time scale of the conformational lattice change that causes the EB binding affinity to change. This most likely reflects more or less directly the time scale of GTP hydrolysis, with the caveat that we do not know how fast phosphate is released.

– The spatially resolved EB binding analysis was a strength of the paper for me. I think it also has the potential to be confusing to people, because the language/fits imply a series of discrete 'states' but it seems these states would have to be heterogenous in terms of nucleotide state. A few more sentences about this might be useful.

We have added a discussion of this issue to our expanded Discussion.

– The supplemental discussion felt looser than the rest of the paper. In particular, while the authors ascribe various deficiencies to nucleotide analogs, they do not seem to consider the possibility that the mutation(s) they made might also perturb the structure. This criticism applies to the main text also. Some version of the discussion of induced fit should probably be incorporated into the main text.

We have re-written the manuscript in a longer format, with a separate Introduction and Discussion. To address the valid concern of the mutation potentially perturbing the microtubule structure, we performed an additional experiment. We removed GTP from purified E254A tubulin by buffer exchange and grew E254A microtubules in the presence of GMPCPP. These microtubules display strongly reduced EB binding, clearly demonstrating that the EB binding affinity, and hence the conformation of the microtubule, is indeed determined by the presence of the nucleotide and not by the E254A mutation itself (Figure 2—figure supplement 1D).

Reviewer #3:

The study by Roostalu et al. addresses the fundamental question of how GTP-tubulin at the growing tip of a microtubule can be detected by other proteins. The present study examines the binding of the major plus end tip tracking protein, EB3, and its ability to specifically recognize the GTP form of tubulin. In recent years, it has been controversial as to whether tip trackers are recognizing GTP or possibly another nucleotide/conformational state, such as GDP-Pi-tubulin. Here, it is shown, using tubulin mutants that either fail to hydrolyze or hydrolyze slowly, that EB3 specifically recognizes the GTP form of tubulin. This is an important finding for the field, one that is achieved through elegant experimental studies of mutant tubulin and careful quantitative analysis, for which the authors are to be commended. However, in the final analysis, the authors invoke a GDP-Pi intermediate state, without strong evidence to justify it. Rather, it seems possible that a simpler alternative explanation that only assumes GTP and GDP states, as suggested by their data, is not ruled out. Thus, I am concerned that the study, while making an important contribution to the field, may be misleading in its final interpretation.

We thank this reviewer for the overall positive assessment of the manuscript and the acknowledgement of the importance of our findings. We believe that some of our statements or the final scheme might have caused a misunderstanding. We did not intend to directly draw conclusions about the GDP-Pi intermediate state from our data as presented here but simply aimed to put our new findings about the GTP hydrolysis into the general context of the current knowledge of different nucleotide states at the microtubule end. We have changed or presentation, expanding our Discussion to clarify this point.

Major comments:

1) Conclusion of a GDP-Pi state based on non-exponentially distributed dwell times is problematic.

a) Figure 4C. The spatially resolved dwell time is interesting, but the two positions that are farther away from the tip appear that they may be bi-exponential. It seems the non-exponential distribution might be expected as the k_off jumps when the hydrolysis occurs, which would give spatially varying dwell times and nonexponential distributions as the hydrolysis can occur at random (Poisson process) during the observation time. So it still seems that the data could be consistent with a simple GTP-GDP model, with no need for an additional third state of GDP-Pi.

We agree. We do not intend to conclude that our data are evidence for having detected a GDP-Pi state. We can however also not exclude that a GDP-Pi state contributes to the observed affinity gradient. We believe that the longer version of our revised manuscript helps to prevent this misunderstanding.

b) Figure 4J is unconvincing as the simpler model of only two nucleotide states (GTP and GDP) has not been ruled out. To rule this out it will be necessary to do model-convolution to make it convincing that the analysis method is not yielding spurious results. Even then, the EB binding could be dependent on the local neighborhood of nucleotide state (see Seetapun et al., 2012, Figure 2—figure supplement 1 for conformational spread modeling), which seems reasonable since the binding site is at the interface between tubulins. Overall, model-convolution on the microtubule addition-loss-hydrolysis and EB binding-unbinding to assess the model is needed to rule out the simpler GTP-GDP model. Even then the argument for a GDP-Pi state is not compelling.

We agree also here. Our schematic figure might have been misleading. We have relabeled the scheme, indicating that the colour gradient is indicative of an EB affinity gradient and not intended to make a statement about a GDP-Pi state.

2) The values for KDs of 8 nM and 40 nM for EB3 binding to GTP- and GDP-tubulin, respectively, seem very strong compared to Seetapun et al. values of 3.8 µM and 55 µM, respectively, in vivo. Why is this, and are the in vitro results informative of the tip tracking in living cells?

The Kd values we have measured for EB3 binding are in the usual range of Kd values measured in in vitro experiments, here at the lower end of the spectrum, because we used a lower ionic strength buffer compared to previous experiments. Why the Kd values for binding the growing microtubule end and the GDP lattice as inferred from fluorescence intensity measurements in cells at a single cellular EB1 concentration are different is currently unclear. Since the in vitro situation is simpler, the EB occupancy on the microtubule can be measured at different EB concentrations, resulting in a binding curve that gives confidence in the obtained Kd value. Independent of the discrepancy between the absolute values of the Kd measurements in vitro and in cells, all measurements agree in that the affinity for the growing microtubule end and – as for the first time measured here – for the GTP lattice is higher than for the GDP lattice. In other words, relative comparisons within one system lead to the similar conclusions.

3) How big is the cap with wildtype tubulin in vitro? Note: "cap size" is in the title, but it is not estimated, despite a lot of nice quantitation. Previous estimates put it at ~40 in vitro (at 5 µM wildtype tubulin) and ~750 in vivo (Schek et al., 2006; Seetapun et al., 2012), a disparity that is largely explained by the relative disparity in the net growth rates under these two conditions. These papers should be cited, as previous estimates of cap size. (Note: need to account for tubulin concentration, e.g. 19 µM wildtype tubulin in Figure 3J vs. 5 µM in Schek et al. and 7 µM in. Seetapun et al.).

The cap size was stated in our Figure 3J in units of length for wildtype and E254D microtubules. The spatial decay rate constants are 1/220 nm and 1/690 nm, respectively. The main point made here was that the cap is elongated when GTP hydrolysis is impaired by the E254D mutation. As suggested by the reviewer, the cap lengths for wildtype and E254D microtubules can also be expressed in numbers of tubulin forming the cap (assuming 13pf microtubules, 8 nm tubulin length and a mono-exponential distribution of cap tubulins): 358 and 1120 tubulins, respectively.

Comparing the cap size measured in vitro for wildtype microtubules with the size of 750 tubulins forming the cap of microtubules in growing in cells, one finds that the cap in cells was 2x larger than it is here in vitro for microtubules growing about 3x faster in cells (156 nm/s vs. 57nm/s). Given that cap length can vary easily by a factor of 2 depending on buffer conditions and given that not the same methods for the determination of the decay rate constants were used in our and the Seetapun study, we can conclude that there is good agreement between cap lengths measured in vitro and in cells, as we have already concluded previously citing the Seetapun et al. paper (Maurer et al., 2014).

In Schek et al., comet lengths were not determined. Therefore, a similar comparison is not possible. Schek et al. however reported transient shrinkage episodes during growth corresponding to up to 40 tubulins per shrinkage events which suggested the necessity for long GTP caps as they are observed using EBs as cap markers. We acknowledged this previously by citing the Schek et al. paper (Duellberg et al., 2016). We have included these citations also in the current manuscript.

Author details

1. Johanna Roostalu

   The Francis Crick Institute, London, United Kingdom

   Present address

   The Wellcome Trust, London, United Knigdom

   Contribution

   Conceptualization, Resources, Formal analysis, Funding acquisition, Validation, Investigation, Visualization, Methodology, Project administration, Supervision

   For correspondence

   johanna@roostalu.info

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-6757-0468

2. Claire Thomas

   The Francis Crick Institute, London, United Kingdom

   Contribution

   Conceptualization, Resources, Validation, Investigation, Methodology

   Contributed equally with

   Nicholas Ian Cade

   Competing interests

   No competing interests declared

3. Nicholas Ian Cade

   The Francis Crick Institute, London, United Kingdom

   Contribution

   Conceptualization, Resources, Software, Formal analysis, Validation, Investigation, Visualization, Methodology

   Contributed equally with

   Claire Thomas

   Competing interests

   No competing interests declared

4. Simone Kunzelmann

   The Francis Crick Institute, London, United Kingdom

   Contribution

   Investigation, Methodology

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-2678-0549

5. Ian A Taylor

   The Francis Crick Institute, London, United Kingdom

   Contribution

   Investigation, Methodology

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-6763-3852

6. Thomas Surrey

   1. The Francis Crick Institute, London, United Kingdom
   2. Centre for Genomic Regulation, Barcelona Institute of Science and Technology, Barcelona, Spain
   3. ICREA, Barcelona, Spain

   Contribution

   Conceptualization, Resources, Supervision, Funding acquisition, Validation, Project administration

   For correspondence

   thomas.surrey@crg.eu

   Competing interests

   Reviewing Editor, eLife

   "This ORCID iD identifies the author of this article:" 0000-0001-9082-1870

• 8,502

  Page views

• 694

  Downloads

• 46

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.