Two-step interphase microtubule disassembly aids spindle morphogenesis

Background Entry into mitosis triggers profound changes in cell shape and cytoskeletal organisation. Here, by studying microtubule remodelling in human flat mitotic cells, we identify a two-step process of interphase microtubule disassembly. Results First, a microtubule-stabilising protein, Ensconsin/MAP7, is inactivated in prophase as a consequence of its phosphorylation downstream of Cdk1/cyclin B. This leads to a reduction in interphase microtubule stability that may help to fuel the growth of centrosomally nucleated microtubules. The peripheral interphase microtubules that remain are then rapidly lost as the concentration of tubulin heterodimers falls following dissolution of the nuclear compartment boundary. Finally, we show that a failure to destabilise microtubules in prophase leads to the formation of microtubule clumps, which interfere with spindle assembly. Conclusions This analysis highlights the importance of the step-wise remodelling of the microtubule cytoskeleton and the significance of permeabilisation of the nuclear envelope in coordinating the changes in cellular organisation and biochemistry that accompany mitotic entry. Electronic supplementary material The online version of this article 10.1186/s12915-017-0478-z) contains supplementary material, which is available to authorized users.


Introduction 46
The goal of mitosis is the equal segregation of genetic material into two daughter 47 cells. To achieve this, animal cells undergo profound changes in cell organisation.

48
Cells round up, chromosomes condense, and the permeability of the nuclear 49 envelope increases (a process we term NEP), leading to mixing of nucleoplasm and 50 cytoplasm. At the same time, the array of long interphase microtubules is replaced by 51 a population of short and highly dynamic centrosome-nucleated microtubules, which 52 go on to form the mitotic spindle -the structure responsible for chromosome

132
To gain a quantitative measure of changes in microtubule polymer levels during entry 133 into mitosis, we followed changes in the intensity of the mEGFP-a-tubulin polymer 134 signal at centrosomes, using the rise in nuclear mEGFP-a-tubulin as a marker of

154
Based on these results, we conclude that the microtubule cytoskeleton is reorganised 155 in two discrete steps at mitotic entry. The first step, during prophase, is characterized 156 by a slow partial depolymerisation of interphase microtubules at the cell periphery.  Step 1: Cdk1/CyclinB dependent removal of Ensconsin from microtubules 167 triggers non-centrosomal microtubule depolymerisation during prophase.

168
To elucidate the molecular mechanisms that govern microtubule disassembly at the 169 entry of mitosis, we focused first on events during prophase. In order to test whether 170 the loss of interphase microtubules during this period is an indirect consequence of 171 the growth of centrosomal microtubules, e.g. via competition for a common tubulin      previously identified as sites of phosphorylation in mitotic cells using mass 232 spectrometry [47,48] (just one of these sites was found to be phosphorylated in both 233 mitosis and G1 [47]). In addition, we identified four potential Nek2 sites in the region.

234
To determine their function, we mutated these ten putative mitotic kinase sites in the 235 context of EMTB-mCherry to generate non-phosphorylatable (here called A-EMTB-236 mCherry) and phospho-mimetic variants (here called, E-EMTB-mCherry). Constructs Next to test whether this phospho-regulation has an impact on interphase 249 microtubule disassembly we transfected a HeLa GFP-a-tubulin stable cell line 250 expressing Rap1* with either a wildtype or an A-mutant version of the EMTB-251 mCherry construct. We then monitored the remodelling of the microtubule 252 cytoskeletal as cells entered mitosis live. Strikingly, the continued association of 253 mutant A-EMTB-mCherry protein with interphase microtubules was sufficient to 254 increase their stability; so that many now persisted into prometaphase (Figure 2 H

357
In this study, we explore how interphase microtubule disassembly is influenced by

377
The mechanism contributing to interphase microtubule disassembly 378 Our analysis points to there being two distinct processes at play. First, the rise of 379 Cdk1/CyclinB activity that drives entry into mitosis induces the phosphorylation of a 380 microtubule stabilizing protein, Ensconsin. This reduces its affinity for microtubules 381 [39,43], making interphase microtubules susceptible to subsequent disassembly.

382
Although these data suggest a specific role for Cdk1/CyclinB in this regulation, the 383 experiments performed do not exclude a role for other mitotic kinases (including 384 Nek2). In addition, even though our data point to the importance of the timely 385 removal of Ensconsin from interphase microtubules, it is likely that the 386 phosphorylation of other MAPs by mitotic kinases also contributes to this process.

387
Indeed, previous work has suggested that the phosphorylation MAP4 by Cdk1 388 14 reduces its ability to stabilize microtubules [59,60]. Moreover, Plk1, another major 389 mitotic kinase, has been shown to stimulate microtubule polymerization activity of 390 MCAK, a key regulator of microtubule dynamics [61,62]. Thus, the activation of 391 CDK1/CyclinB is likely to trigger the remodelling of the microtubule cytoskeleton 392 through several parallel processes. While this is the case, our data show that while 393 the rise in mitotic kinase activity sets the stage for the depolymerisation of 394 microtubules following the loss of nuclear/cytoplasmic compartment boundary, it is 395 not in itself sufficient for the complete disassembly of interphase microtubules.

615
Graphs were produced and statistical analysis (

822
Microtubule-associated proteins control the kinetics of microtubule nucleation.

Relative time to NEP (min)
Step 1 Step 2

Figure1. Disassembly of interphase microtubules begins prior to NEP and is accelerated at NEP.
A) Representative time-lapse confocal images (x-y maximum projection) of a HeLa cell stably expressing H2B-mRFP (to visualize chromosomes) and mEGFP-α-tubulin (to visualize microtubules and NEP), and transiently overexpressing Rap1* (to keep cell flat as it enters mitosis). Inserts show regions zoomed in B and C. B) Higher magnification (sum projection of sections of mEGFP-αtubulin around the centrosome, pseudo-color, spectra LUT) of boxed region 2 indicated in A showing how mEGFP-α-tubulin levels at the centrosome rise prior to NEP. Inserts indicate regions used for quantifications: green (centrosomal microtubules), red (nuclear tubulin). C) Higher magnification (maximum projection of basal sections of mEGFP-α-tubulin, inverted grayscale) of region 1 above showing that non-centrosomal microtubule disassembly is triggered before NEP and accelerates during loss of the nuclear/ cytoplasmic compartment boundary. Insert indicates region used for quantifications. D) Changes in median centrosomal and non-centrosomal microtubule intensity relative to NEP for H2B-mRFP mEGFP-α-tubulin HeLa cell transiently overexpressing Rap1* (shown in A-C, left), and for 5 equivalent cells from 2 independent experiments (right). Median intensity of mEGFP-α-tubulin signals was calculated within a 15X15 pixel circle around the centrosome (green line, as indicated in B), a 10X10 pixel box within the nucleus (red, as indicated in B), and a 30X30 pixel box at the cell periphery (blue, as indicated in C). Time point 0 represents NEP. Graph shows means and SD. E) Changes in non-centrosomal microtubule levels relative to NEP. Measurements show median of mEGFP-α-tubulin signal in a 30X30 pixel box at two locations at the periphery of a cell as shown in Figure S1A at 30, 20, 2 min before NEP and 4, 6 min after NEP in H2B-mRFP mEGFP-α-tubulin HeLa stable cell line transiently overexpressing Rap1* (13 cells (include 5 cells from 1D), 4 independent experiments). Repeated Measures ANOVA, Tukey's multiple comparisons test with a single pooled variance, **** P<0.0001. Scale bars represent 10μm.  hite arrows indicate to microtubule clumps formed due to a failure to remove Ensconsin from microtubules before NEP. Black arrows indicate to interphase microtubules just before or after NEP (1 cells t-EMTB-mcherry, 15 cells A-EMTB-mcherry, 5 independentexperiments). In overlay images, signal intensities were adjusted to remove cytoplasmic background signal. Scale bars represent 10μm.   A) Representative time-lapse confocal images of a HeLa cell during mitotic entry stably expressing H2B-mRFP (was not imaged) and mEGFP-α-tubulin, and transiently overexpressing Rap1* treated with STLC. Upper panel: upper single section (inverted grayscale) of mEGFP-α-tubulin to visualize NEP, middle panel: x-y maximum projection images of mEGFP-α-tubulin (Inverted grayscale) to visualize microtubules, lower panel: sum projection of sections of mEGFP-α-tubulin around the centrosome (pseudo-color, spectra LUT) to visualize centrosomal microtubules showing the NEP is correlated with transient decrease in mEGFP-α-tubulin intensity at centrosomes. B) Changes in centrosomal microtubule levels (circles) relative to NEP (nuclear tubulin, triangles) for one cell (shown in A, upper panel) and for equivalent 4 cells from 2 independent experiments (lower panel) measured similar to that described in Figure 1 D (see Materials and Methods). C) Representative time-lapse confocal images of a HeLa cell expressing MS2-mCherry-NLS (nuclear marker) and Lifeact-GFP during mitotic entry. D) Quantification of volume occupied by MS2-mCherry-NLS signal in HeLa cells similar to 4 C ( cells, 2 independent experiments) entering mitosis. Graph shows means and SD. E) uantification of MS2-mCherry-NLS intensity. Median of MS2-mCherry-NLS signal is shown for 20X20 pixel box at two locations in the nucleus per cell at 30,10, 4 min before NEP and 2, 4,10 min after NEP ( cells, 2 independent experiments, the same cells as in 4 E). Repeated Measures ANOVA with the Greenhouse-Geisser correction, Tukey's multiple comparisons test with individual variances computed for each comparison, **** P<0.0001. F) Representative time-lapse confocal images of a HeLa cell stably expressing H2B-mRFP (shown in the upper panel at -30 min and 6 min) and mEGFP-α-tubulin during mitotic entry, following treatment with Nocodazole in non-adherent chambers (upper panel: x-y maximum projection: lower panel single section, pseudocolor, spectra LUT) to show cytoplasmic tubulin dilution at NEP. G) Quantification of median mEGFP-α-tubulin signal in 6X6 pixel box in nucleus and at two locations at the periphery of cells as in 4F ( cells, 1 independent experiment) entering mitosis in the non-adherent chambers filmed at high resolution using spinning disc confocal microscope. H) uantification of median mEGFP-α-tubulin signal in 4X4 pixel box at four locations at the periphery of the cells similar to 4 F (20 cells, 2 independent experiments) at 2 , 6, 3 min before NEP and 3, 6 min after the NEP in non-adherent chambers filmed at low resolution with wide-field microscope. Repeated Measures ANOVA with the Greenhouse-Geisser correction, Tukey's multiple comparisons test, with individual variances computed for each comparison, **** P<0.0001. Scale bars represent 10μm. A) Representative time-lapse confocal images (x-y maximum projection, lower panel: pseudo-color, spectra LUT) of HeLa cells stably expressing H2B-mRFP (not imaged) and mEGFP-α-tubulin showing changes in spindle mEGFP-α-tubulin intensity and spindle volume upon hypo-or hyper-osmotic shock relative to a control. Scale bar represents 10μm. B) uantification of changes in spindle mEGFP-α-tubulin intensity induced by osmotic shock. Mean intensity of spindle mEGFP-α-tubulin intensity was measured by rendering mitotic spindles in 3D using Imaris software before and after control (blue, 12 cells), hypo (red, 14 cells) or hyper (green, 10 cells) osmotic shock treatment (2 independent experiments). Lower panel: comparison between -0.5 min and 2 min relative to osmotic shock treatment. Repeated Measures two-way ANOVA, Dunnett's multiple comparisons test, **** P 0.0001. C) uantification of changes in spindle volume induced by osmotic shock. Spindle volume was measured by rendering mEGFP-αtubulin signal in 3D in Imaris before and after control (blue, 12cells), hypo (red, 14) or hyper (green, 10 cells) osmotic shock treatment (2 independent experiments). Lower panel: comparison between -0.5 min and 2 min relative to osmotic shock treatment. Repeated Measures two-way ANOVA, Dunnett's multiple comparisons test, **** P 0.0001. A) Representative confocal image (x-y maximum projection) of H2B-mRFP (not shown) mEGFP-α-tubulin HeLa cells (inverted grayscale) transiently overexpressing Rap1* entering mitosis (the same cell as in Figure 1) and a control cell that remains in interphase during the course of the movie. Inserts show regions that were used for quantifications in B-D. Scale bar represents 10μm. B) Changes in non-centrosomal microtubule levels in the cells entering mitosis (blue, the same as in Figure 1D) vs. interphase cells (black, 5 cells from 2 independent experiments). Median intensity of mEGFP-α-tubulin signal was measured as in Figure 1 D. Graph shows means and SD. Changes in non-centrosomal microtubule levels in the cells entering mitosis (blue, the same as in Figure 1 E) vs. interphase cells (black, 13 cells from 4 independent experiments). Median (C) and Variance (D) of mEGFP-α-tubulin signal was measured as in Figure 1 E. E) Changes in centrosomal microtubule levels. Mean of α-tubulin-GFP signal at the centrosomes was measured as in Figure 2 B in the cells entering mitosis at 30, 20, 2 min before NEP and 4, 6 min after NEP (13 cells, 4 independent experiments, the same cells as in Figure 1). Repeated Measures ANOVA with the Greenhouse -Geisser correction, Tukey's multiple comparisons test with individual variances computed for each comparison. Shown * P=0.039. A) Changes in non-centrosomal microtubule levels relative to NEP. Variance of α-tubulin-GFP signal measured as described in Figure S1 D for control siRNA (blue, 5 cells) and Cep192 siRNA cells (brown, 5 cells) 2 experiments as in Figure 2 A-C. B) Western blot showing Ensconsin knockdown induced using three different siRNAs targeting Ensconsin. C) Representative confocal images (x-y maximum projection) of fixed HeLa cells treated with control siRNA and three siRNAs targeting Ensconsin. Inserts show regions zoomed in overlays (>10 cells per condition, 1 experiment). D) Representative confocal images (x-y maximum projection) of fixed MCF10A cells stained to show that Ensconsin is removed from microtubules in prophase compared to interphase. Ensconsin in red, αTubulin in green and DAPI in blue. Inserts show regions zoomed in overlays, in which intensities were adjusted to remove cytoplasmic background signal (6 prophase cells, >20 interphase cells,1 experiment). E) Representative confocal images of fixed HeLa cells overexpressing Rap1* in prophase stained to show that the microtubule binding domain of Ensconsin (Wt-EMTB-mCherry) as well as a corresponding phospho-mimetic mutant (E-EMTB-mCherry) are largely cytoplasmic in prophase, whereas the non-phosphorylatable form (A-EMTB-mCherry) localizes to the microtubules. αTubulin in green, Wt-EMTB-mCherry, A-EMTB-mCherry, E-EMTB-mCherry in red. Inserts are zoomed, shown in inverted greyscale or in overlays, where signal intensities were adjusted to remove cytoplasmic background signal (3 cells per condition, 1 experiment). Scale bars represent 10μm.  A) Representative time-lapse confocal images (x-y maximum projection, lower panel: pseudocolor, spectra LUT) of HeLa cells stably expressing H2B-mRFP (was not imaged) and mEGFP-α-tubulin, treated with Nocodazole, to show changes in cell diameter and in mEGFP-α-tubulin intensity before and after hypo-or hyper-osmotic shock treatment relative to control treatment. Scale bar represents 10μm. B) Quantifications of changes in mEGFP-α-tubulin intensity induced by osmotic shock relative to control treatment. Mean intensity of mEGFP-α-tubulin signal was measured in cells before and after control (blue, cells), hypo-(red, cells) or hyper-(green, cells) osmotic shock treatments (2 independent experiments) as described in Materials and Methods. C) uantifications of changes in cell diameter (Ferret s diameter) induced by osmotic shock relative to control treatment. Ferret's diameter was measured as described in Materials and Methods in cells before and after control (blue, cells), hypo-(red, cells) or hyper-(green, cells) osmotic shock treatments (2 independent experiments). Lower panels show comparison between values at -0.5 min and 2 min relative to osmotic shock treatments. Repeated Measures two-way ANOVA, Dunnett's multiple comparisons test, **** P=0.0001